NEMBES-Networked Embedded Systems | University College Cork

Improving network reliability and survivability

The deployment of sensors determines the topology of the network, which will later influence the coverage and the efficiency of the WSN. The robustness of the topology comes from its reliability. The WSN reliability is defined by several parameters, such as the number of neighbors a node has, the distance between each pair of nodes, connectivity, the existence of disjoint paths, etc. Two paths are node-disjoint (respectively link-disjoint) if both of them do not share any nodes (respectively any links), except the source and the sink nodes. Node-disjoint paths are more resilient to failures than link-disjoint paths as they protect against both node and link failures. At the moment, we are investigating the reliability of the network on the presence of node-disjoint paths with length constraints.

Networked Embedded Systems

Members:

NEMBES is a multi-partner interdisciplinary research project funded by the Irish HEA PRTLI-IV programme. Within NEMBES our goal is to improve decision-making in networked embedded systems, combining optimisation expertise of Cork Constraint Computation Centre (4C) and networking expertise of Mobile and Internet Systems Laboratory (MISL). Our current focus is emergency response within buildings, facilitating efficient evacuation and rescue. Below are the topics that we are currently investigated.

In normal mode (no fire), ER-MAC is similar to TDMA. Sensor node can communicate only in its own time-slot and it sleeps in other time with the aim to be highly energy-efficient. In emergency mode (fire), ER-MAC let sensor nodes compete for time-slots based on the time-slot ownership and data priority. ER-MAC has been implemented in both NS2 and Contiki-OS, and it has been evaluated through extensive simulations and experiments on a small testbed, and the results are very promising.

ORMS exploits the emergency responders (mobile sinks) to improve the performance of wireless sensor network, which tends to be highly-loaded and partitioned in emergency situations. We studied the broadcast range of the mobile sink's presence and the way of exploiting the prediction of mobile sink's mobility. ORMS has been implemented in both NS2 and Contiki-OS, and it has been evaluated through extensive simulations and experiments on a small testbed, and the results are very promising.

In opportunistic data collection, data can be collected only when mobile node and sensor node are close and they know the presence of each other. Hence, contact probing mechanism must be well designed to maximize the probed contact capacity under the duty-cycle constraint of sensor nodes. SNIP, a sensor node initiated probing mechanism, has been proposed by us. SNIP has been implemented in Contiki-OS, extensive simulations have been carried out with Cooja, and the results are very promising.

Furthermore, sensor nodes tend to be visited by mobile node with different probability. It is preferable if the sensed data could be pre-forwarded to nodes that are visited more frequently. Heuristic algorithms have been designed for data pre-forwarding with the aim to improve the performance (throughput, latency) of data collection while maintaining low protocol overhead. These heuristics will be implemented in Contiki-OS and evaluated through simulations (Cooja) and experiments (a small testbed).

ORMS exploits the emergency responders (mobile sinks) to improve the performance of wireless sensor network, which tends to be highly-loaded and partitioned in emergency situations. We studied the broadcast range of the mobile sink's presence and the way of exploiting the prediction of mobile sink's mobility. ORMS has been implemented in both NS2 and Contiki-OS, and it has been evaluated through extensive simulations and experiments on a small testbed, and the results are very promising.

A Real-Time and Robust Routing Protocol for Building Fire Emergency Applications Using Wireless Sensor Networks

For building fire emergency, the key requirement is to deliver messages in real-time and with a high probability of success and it is very challenging for wireless sensor networks. Based on this observation, Real-time and Robust Routing in Fire (RTRR) is proposed by us. To satisfy the above requirement, the following techniques are adopted by RTRR.

In RTRR, the delay from each node to its nearest sink is estimated to guide a real-time delivery.
RTRR also tracks the status of nodes and link valid time in fire, allowing traffic to avoid nodes that are in danger according to fire spreading
Adaptive transmission power is adopted to avoid routing holes (caused by nodes that have failed) or seek real-time and valid paths in fire situations.

Opportunistic Routing Protocol

When fire occurs in a building, sensor nodes tend to generate more data and the network tends to be partitioned since sensor nodes may be burnt. It is very challenging to transmit the data to the sink, at which the evacuation plan will be made. Based on this observation, we propose to let emergency responders (firefighters) collect data from sensor nodes and ORMS, an opportunistic routing protocol, is designed for such kinds of emergency response applications.

Opportunistic Data Collection

With the maturing of wireless sensor networks, we can expect that some large-scale networks will be deployed for providing long-term services. Under this scenario, it becomes very attractive to let the passengers opportunistically collect data from sensor nodes. Through exploitation of the un-controllable but free mobility, the cost of data collection can be reduced significantly.

Real-time Evacuation Simulator

EvacSim is a multi-agent based Emergency Evacuation simulator. The simulator is designed to evaluate evacuation scenarios and egress plans in case of emergency in a building and consists of:

Building model (derived from an Industry Foundation Classes building definition)
Building Occupant Agents
Simulated Embedded Systems (sensors, signposting etc)

EvacSim operates in real time simulating occupant behaviour during a fire event. The occupant agents feature flocking and crowding behaviour, plan their route to exit the building and react to events in the building such as following exit signs or avoiding smoke-filled zones. Simulated WSN components include motion detectors, smoke detectors, actuated exit signs and targetted direction (e.g. individual direction via smartphone). The simulated sensor components can provide a sensor "perspective" of the evacuation progress which corresponds to the information that would be available from sensors during a real emergency. This information can be used with an evacuation planning algorithm (e.g. maximum flow evacuation routing) to direct actuation of networked signposting in the simulated building.

The simulated evacuation can be used to evaluate evacuation planning algorithms on metrics such as time to clear building, average evacuation time per individual and worst case evacuation time.

Emergency Response MAC

In the context of building fire detection and evacuation application, MAC protocol used by sensor nodes must be highly energy-efficient when there is no fire and it should be able to transmit the important and high-rate data generated by sensor nodes that detected the fire. ER-MAC, a hybrid MAC protocol, has been proposed by us to satisfy the above requirements.

Optimal Evacuation Guidance

By combining the presence detection and hazard detection based on wireless sensor networks with localized actuation of evacuation direction signs, it is possible to guide the movement of people during the evacuation process. An important issue is to design the guidance algorithms that based on the estimated locations of people and the estimated spread of the hazard (such as fire) execute the evacuation so that it is carried out in the safest and fastest possible way. The standard approaches to egress optimisation (estimating the minimal evacuation time) are based on the well developed theory of dynamic network flows. We however, also require that the algorithms are adaptive, i.e. they can readjust to the changes during the evacuation, and distributed, i.e. local sensor/actuation nodes can help with the evacuation process even if the part of the sensing network infrastructure is damaged.

A Dynamic Model for Fire Emergency Evacuation Based on Wireless Sensor Networks

During building fire emergency, the environment tends to change with time and it is very challenging to generate the evacuation plan. Based on wireless sensor network, a dynamic model is proposed for fire emergency. According to the data reported from sensor nodes, this dynamic model provides estimated information about the dynamicity of the fire hazard over time in the building environment. The model then generates a set dynamic navigation weights C(u,v,t) representing the time taken to walk between two adjacent locations u, v at the time t. Based on these elements two types of dynamic navigation paths are introduced within the building environment. Firstly, the dynamic shortest paths are considered to be used by well-able evacuees towards the exit or by the fire-fighters to navigate in the building. The second type of path uses the concept of safety which represents the maximum time one can safely delay at the nodes. These dynamic safety paths can be used in evacuation by evacuees with disability of by fire-fighters assisting injured evacuees.